Comprehensive primary particulate organic characterization of vehicular exhaust emissions in France

Atmospheric Environment 43 (2009) 6190–6198

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Comprehensive primary particulate organic characterization of vehicular exhaust emissions in France Imad El Haddad a, *, Nicolas Marchand a, *, Julien Dron a, Brice Temime-Roussel a, Etienne Quivet a, Henri Wortham a, Jean Luc Jaffrezo b, Christine Baduel b, Didier Voisin b, Jean Luc Besombes c, Gregory Gille d a

Universite´s d’Aix-Marseille I, II et III – CNRS, UMR 6264: Laboratoire Chimie Provence, Equipe: Intrumentation et Re´activite´ Atmosphe´rique 3 place Victor Hugo, 13331 Marseille Cedex 3, France Universite´ Joseph Fourier – Grenoble 1 – CNRS, UMR 5183, Laboratoire de Glaciologie et Ge´ophysique de l’Environnement Rue Molie`re, BP 96, 38 402 St Martin d’He`res Cedex, France c Laboratoire de Chimie Mole´culaire et Environnement, Polytech’Savoie- Universite´ de Savoie, Campus Scientifique, 73376 Le Bourget du Lac Cedex, France d Regional Network for Air Quality Monitoring (ATMO-PACA), 146 rue Paradis 13006 Marseille, France b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 8 April 2009 Received in revised form 2 September 2009 Accepted 2 September 2009

A study to characterize primary particulate matter (PM2.5 and PM10) from the French vehicular fleet was conducted during winter 2008, in a tunnel in Marseille, France. The carbonaceous fraction represents 70% of the aerosol mass and elemental carbon fraction (EC) represent 60% of the carbonaceous fraction. The organic carbon OC was characterized in term of its water soluble fraction, functionalization rate and HULIS content. Seventy trace organic compounds including alkanes, polycyclic aromatic hydrocarbons (PAH), petroleum biomarkers and carboxylic acids were also quantified, in order to determine an organic emission profile for chemical mass balance modeling studies. Such source profiles were still missing in Europe and particularly in France. The profile obtained in this study is consistent with profiles determined in tunnel or dynamometer studies performed in other countries during the last ten years. These results suggest that organic compounds profiles from vehicular exhaust emissions are not significantly influenced by the geographic area and are thus suitable for use in aerosol source apportionment modeling applied across extensive regions. The chemical profile determined here is very similar to those obtained for diesel emissions with high concentrations of EC relative to OC (EC/OC ¼ 1.8) and low concentrations of the higher molecular weight PAH. These results are consistent with the high proportion of diesel vehicles in the French fleet (49%). Ó 2009 Elsevier Ltd. All rights reserved.

Keywords: Organic aerosol Vehicular emissions Organic markers Functional analyses Tunnel experiment Marseille

1. Introduction Vehicular emissions are one of the major primary fine particulate matter (PM2.5) sources. It accounts for approximately 15–50% of total fine aerosol mass in urban areas (Sheesley et al., 2007 and references therein). Vehicular emissions are also of specific interest, since the particles emitted are essentially in the size range of 20–120 nm (Jamriska et al., 2004) and potentially contain toxic components, such as polycyclic aromatic compounds (PAH) and trace metallic elements (Schauer et al., 1999, 2002; Ntziachristos et al., 2007), which relate them to acute and chronic cardiovascular and respiratory outcomes (Grahame and Schlesinger, 2007).

* Corresponding authors. Tel.: +33 491108512; fax: +33 491106377. E-mail addresses: [email protected] (I. El Haddad), Nicolas. [email protected] (N. Marchand). 1352-2310/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2009.09.001

Primary particles from vehicular exhaust emissions are mainly composed of organic carbon OC and elemental carbon EC, with lesser amounts of trace metals and inorganic ions (Schauer et al., 1999, 2002). Many of the particulate OC compounds, commonly referred to as molecular markers, can be used to estimate the contribution of primary vehicular emissions in ambient atmosphere, using Chemical Mass balance (CMB) receptor-based model. This modeling approach requires a prior knowledge of emission profiles in order to apportion the sources (Schauer et al., 1996). The vehicle exhausts profiles are often based on chassis dynamometer studies (Rogge et al., 1993a; Schauer et al., 1999, 2002; Zielinska et al., 2004). Such studies offer the advantage of controlled testing conditions, which in turn enable the evaluation of the effect of different engine designs, fuel composition, or driving conditions on pollutant emissions. However, these tests may not provide a good representation of the entire vehicle fleet because of the large variations in engine types and the maintenance histories of actual

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

on-road vehicles. In addition these studies do not account for nonexhaust emissions from tire wear, break-wear, and re-suspended road dust which can be also a significant contributor to PM concentrations (Querol et al., 2004). Vehicular emission chemical profiles can be determined using an alternative approach, based on highway tunnels and near-roadside measurements (Miguel et al., 1998; Wang et al., 2006; Ntziachristos et al., 2007). Roadway tunnel studies are well suited to determine an average vehicular exhaust profile representative of total vehicular fleet at operating conditions with little interference from other sources. Furthermore, tunnel studies provide samples that are representative of exhaust and non-exhaust emissions at atmospheric conditions (temperature, relative humidity). The source profiles of organic compounds from vehicle emissions are relatively well established in the US, and have been applied successfully to apportion ambient PM primary sources using CMB modeling (Schauer et al., 1996; Ke et al., 2007; Sheesley et al., 2007). However, data sets available for European and particularly for French vehicle particle emissions are scarce and a detailed particulate chemical composition had never been reported before. In an European context of reduction of PM2.5 concentrations (target value of 20 mg m3 in urban areas for 2015) (2008/50/EC, 2008), the knowledge of the chemical fingerprint of particles emitted by vehicular traffic in Europe is required in order to evaluate more accurately the influence of this primary source to the total atmospheric load of PM. This paper presents results from a sampling campaign carried out in a roadway tunnel, in Marseille, France. The detailed chemical composition of primary vehiclederived particles (PM2.5 and PM10) was determined, with a special focus on organic speciation. 2. Method 2.1. Measurement site and vehicle fleet characteristics Field measurement was conducted in a roadway tunnel, in Marseille, France. The tunnel is 2455 m long, and consists of two unidirectional superposed roadways that include 2 lanes each. The measurement site was located inside the tunnel approximately 200 m from the exit (according to the direction of the traffic) and 3 m away from the vehicle traffic. The vehicle speed is limited to 50 km h1 and the air flow is likely to be dominated by the vehicle motion (piston effect), since the traffic is unidirectional. Hence, the sampling site is mainly influenced by the vehicles emission along the tunnel. In the tunnel, the heavy duty trucks are banned and the traffic flow is composed of light-duty vehicles, including motorcycles. Total traffic, in the tunnel tube under study, counts between 20,000 and 25,000 vehicles per day, with significant diurnal variations. The average daytime (resp., night-time) traffic density was about 1700 (resp., 250) vehicles per hour (Fig. 1). No traffic jam occurred during the sampling period. The vehicles passing the tunnel were not tracked to perform their classification by engine technologies. However, the global French vehicular distribution (excluding heavy duty trucks) comprises 30 million vehicles (0.49 vehicles per person) with an average age of 7.7 years, of which 49% are diesel powered (Arthaut, 2005). In contrast, in the US, the majority of the on-road fleet are attributed to light-duty vehicles, which are fuelled almost entirely by gasoline (Gertler, 2005). 2.2. Pollutant measurement and sample collection The study took place on February 19th and 20th, 2008. The concentrations of the gaseous (CO and NOX) and particulate (PM10 and PM2.5) pollutants were monitored with Environment SA and TEOM (Thermo Scientific) analyzers, respectively. The aerosol

6191

number size distribution was monitored in the range 11.1–1083 nm with a Scanning Mobility Particle Sizer (SMPS, L-DMA, CPC5403, GRIMM). PM collection was performed with a high volume sampler (DA-80 DIGITEL), operating at 30 m3 h1. The samples were collected onto 150 mm diameter quartz filters (Whatman, Q-MA), previously fired for 2 h at 500  C. Sampling took place between 7 a.m. and 8 p.m. (local time), capturing the rush hours events. On February 19th, 12 hourly PM2.5 samples were collected, while on February 20th, 9 half-hour PM10 samples were collected. Samples were stored at 18  C in aluminium foil, sealed in polyethylene bags until analysis. Just before the analyses, 2 PM2.5 and 2 PM10 composite samples were prepared, by combining the exposed filters, chosen in alternate order. Two field blank filter composites were also prepared following the same procedure. High volume sampling induces well-known collection artifacts (adsorption/volatilisation and chemical transformation onto the filter during sampling) (Mader and Pankow, 2001; Sihabut et al., 2005; Goriaux et al., 2006). Although these artifacts can significantly modify the concentrations of some organic compound, high volume samplers are still widely used particularly for reliability reasons. These artifacts are highly dependant on the sampling conditions (temperature, oxidant concentrations, air masses ages and sampling time), and corrections are still difficult to apply. Therefore no corrections have been applied in this study. However, organic compounds for which sampling artifacts could induce errors higher than 30% are quoted in Tables 2–4, according to Sihabut et al. (2005) for adsorption artifacts and Goriaux et al. (2006) for chemical transformation artifacts. Note that other semivolatile compounds (i.e. lower molecular weight (MW) n-alkanes (
6192

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

Fig. 1. General characteristics of the atmospheric environment inside the tunnel. (a) Traffic density (NV), particles number concentrations (NP) and pollutants (CO, NOX and PM) diurnal variation during the measurement period. (b) Daytime aerosol number size distribution. (c) Total particle number concentration (NP) vs traffic density (NV).

The procedures used for extraction, derivatisation and analyses of the organic compounds are based on the Desert Research Institute standard operating procedure for the analysis of semi-volatile organic compounds (DRI, 2003) and are detailed in the SI. All samples were analyzed using a Trace GC 2000 chromatograph coupled to an ion trap mass spectrometer (Polaris Q, Thermo Scientific) and equipped with a TR-5MS capillary column (Thermo Scientific, 30 m  0.25 mm i.d.  0.25 mm film thickness). In addition, PAH were analyzed by reversed phase high performance liquid chromatography coupled to a UV fluorescence detection following the method described in Besombes et al. (2001). The quantification of the organic compounds was performed using authentic standards (Tables 2–4). Compounds for which no authentic standard were available were quantified using the response factor of compounds with analogous chemical structures (Tables 2–4).

concentrations were 4600 ppb, 2200 ppb, and 320 mg m3, respectively, while during the night their average concentrations decreased to 2300 ppb, 1100 ppb, and 150 mg m3, respectively (Fig. 1a). The aerosol number size distribution within the tunnel showed a single mode at 50–60 nm during the whole sampling period (Fig. 1b), characteristic of primary particles originating from fuel combustion, mainly diesel soot (Jamriska et al., 2004). Total particle number concentration was highly correlated with the traffic intensity (NV), confirming that submicrometer particles originated primarily from the primary vehicular emissions (Fig. 1c). No control samples have been collected outside the tunnel during the observation period. However, background influence inside the tunnel can be estimated considering PM2.5 and PM10 concentrations during the sampling period. The average PM10 concentration during the collection period (daytime) at a traffic station upwind from the tunnel was 28.1 mg m3. Considering this value and a transmission efficiency of background particles inside the tunnel equal to one, maximal PM10 background contribution of about 9% can be estimated. In these conditions and considering an OC/PM of 1/3 (typical value for urban atmosphere; Jaffrezo et al., 2005) the influence of background to OC inside the tunnel can be estimated to be about 14% (Table 1; PM10-NIOSH). Note that these values denote a highest contribution estimation of the non traffic related PM and OC from the background given that this site is highly impacted by vehicular emissions. Otherwise, PM2.5 and PM10 average concentrations in urban background stations over Marseille during the collection period were 13.2 and 17.6 mg m3, respectively. Considering these values and the same assumptions as above, the influence of the background sources to PM and OC can be estimated to be about 5% and 8%, respectively. Consequently, the background highest contribution is estimated to be lower than 9% and 14% for PM and OC, respectively.

3. Results and discussions

3.1. Mass balance of traffic related primary particles

The concentrations of CO, NOX, and PM10 in the tunnel showed a strong diurnal variation. The CO, NOX, and PM10 average daytime

PM2.5 and PM10 composition is detailed in Table 1, and an overall PM2.5 chemical mass balance is presented in Fig. 2. PM2.5 and PM10

2.4. Functional groups analysis The analysis was performed by atmospheric pressure chemical ionisation (APCI)-tandem mass spectrometry (Varian 1200L). This technique enables a quantitative determination of the carboxylic, carbonyls, and nitro functional groups, based on their ability to loose a specific neutral fragment or to produce a characteristic ion in the collision cell (Dron et al., 2007, 2008a,b). This methodology is based either on neutral loss mode for carboxylic acid (NL32 after a BF3/MeOH derivatisation step, Dron et al., 2007) and carbonyl functional group (NL181, after a PFPH derivatisation step, Dron et al., 2008b) either on precursor ion scanning mode for nitro functional group (PAR 46) (Dron et al., 2008a). The analytical procedure is detailed in the SI. 2.5. Organic speciation analysis

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198 Table 1 Mass balance and physico-chemical proprieties of traffic related primary PM2.5 and PM10 (average (min–max)). PM2.5 Particulate matter mass [mg m3] PM 282 (272–293)

PM10 321 (297–345)

3

Carbonaceous fraction [mg m ] OC (EUSSAAR2) 44.5 (34.5–55.6) EC (EUSAAR2) 122 (107–138) OC (NIOSH) 59.2 (45.8–77.5) EC (NIOSH) 112 (101–142) WSOC 9.31 (8.41–10.1) HULIS (WS) 1.71 (1.71–2.28) Major ions [mg m3] Chloride Sulfate Nitrate Sodium Ammonium Potassium Magnesium Calcium

0.221 2.31 2.09 0.504 1.28 0.230 0.281 3.72

(0.19–0.26) (2.15–2.47) (1.91–2.28) (0.43–0.61) (0.98–1.58) (0.180–0.260) (0.270–0.290) (3.20–4.04)

Functional group [nmol m3] R-COOH 34.5 (31.0–37.9) 11.7 (10.7–12.7) R-CO-R0 R-NO2 34.4 (27.6–41.3)

59.3 134 79.3 129 13 2.69 1.79 2.47 4.74 1.56 1.59 0.602 0.570 11.8

(48.3–68.2) (126–159) (63.4–86.6) (103–154) (10.8–15.3) (2.69–3.98) (0.55–3.07) (1.51–3.04) (3.98–5.40) (1.41–2.07) (1.41–1.82) (0.560–0.780) (0.460–0.620) (10.5–14.6)

42.4 (29.1–55.7) 19.8 (17.9–21.7) 39.9 (29.5–50.2)

samples were collected, respectively, during the first day and the second day of the campaign. Between the 2 days, the roadway and wall surfaces of the tunnel has been intensively washed using heavy dedicated engines. This monthly cleaning procedure consisted of a roll-over type wash associated with jet sprays of water. Therefore, re-suspension conditions have been significantly modified between the two sampling periods and the comparisons between PM2.5 and PM10 could not be considered as relevant. Nevertheless, these modifications offer an interesting case study for organic compounds associated with the fine aerosol fraction. The particle mass mainly consisted of carbonaceous matter with smaller contributions of inorganic ions. The NIOSH and EUSAAR2 methods gave similar concentrations of total carbon (Table 1). However, the EC (resp. OC) contribution to the TC measured by EUSAAR2 method was about 18% higher (resp. lower) than the ones measured with the NIOSH method. This results from the lower temperature used for the last OC step in the EUSAAR2 method (Cavalli et al., in press). The results reported below are based on the

Fig. 2. PM2.5 mass balance. OM fraction is calculated from OC values, considering OM/ OC ratio of 1.2 (Turpin and Lim, 2001). OM and EC fractions are determined based on the measurements made by the NIOSH method. The HULIS contribution is determined using OM/OC ratio of 1.81 (Salma et al., 2007).

6193

concentrations determined by the NIOSH method, in order to compare our data with previous measurements (Rogge et al., 1993a; Schauer et al., 1999, 2002; Zielinska et al., 2004) that use the NIOSH method. The contributions of the carbonaceous matter to the PM2.5 and PM10 masses were not significantly different. In both size ranges, the EC accounted for 39–42% of the aerosol mass, whereas the organic matter (OM) contribution was between 25 and 30% (Fig. 2), considering an OM/OC ratio of 1.2 (Turpin and Lim, 2001). The EC/ OC ratio is a useful parameter that can discriminate the emissions from diesel and gasoline engines. The ratios measured for the gasoline engines range typically between 0.1 and 0.4, whereas they range between 1.4 and 2.5 for diesel emissions (Rogge et al., 1993a; Schauer et al., 1999, 2002; Zielinska et al., 2004). The EC/OC ratios measured in this study were between 1.6 and 1.9, implying a contribution well above 50% of diesel exhausts to the PM budget, which is consistent with the high proportion of diesel vehicles in the French fleet and their high emission factors comparing to gasoline vehicles. The total measured inorganic fraction accounted for 4% and 8% of the total PM2.5 and PM10 mass, respectively. Nitrate was the largest anionic constituent, representing 0.8% and 1.4% of the PM2.5 and PM10 mass, respectively. Sulphate was the second largest anionic component, accounting for 0.85% to the PM mass on average in both size ranges. Comparable contributions to the PM were reported for dynamometer chassis studies (Schauer et al., 1999, 2002). The most abundant cationic components were the calcium ions; it accounted for 1.4% and 3.7% of the total PM2.5 and PM10 mass, respectively. Elemental calcium is a marker of lube-oil combustion since it is used as anti-wear, detergent, and stabilizing additive in oils (Ntziachristos et al., 2007). It was measured previously in the vehicular particle emissions (Schauer et al., 1999, 2002), but only in trace amounts (<0.35%) that cannot account for the calcium concentrations measured here. The high abundance of calcium ions can be explained by the contribution of non-tailpipe emission, mainly the re-suspension of the geological materials from the roadway. A significant fraction of the PM mass remained unidentified (20–30%). Trace elements associated with emissions from vehicle exhausts (P, Ca, Fe, Zn, .), vehicle brake abrasion (Cu, Sb, Ba, .) and road dust (Al, Si, .) (Schauer et al., 1999, 2002; Ntziachristos et al., 2007) could potentially explain the unidentified fraction. 3.2. Physico-chemical characteristics of OC The WSOC contributed on average to 20% of OC. Such a WSOC fraction is considerably lower than those previously measured for a biomass burning plumes (71%) (Sannigrahi et al., 2006) or at urban sites in the French Alps valleys (between 55% and 75%) (Jaffrezo et al., 2005). Although these ratios reflect the dominant hydrophobic nature of the OA from primary vehicular emissions, it should be noted that there is still a significant soluble fraction even in these conditions. Background influence cannot be totally neglected for WSOC. The hydrophobic character of vehicles-emitted organic aerosol was also revealed by its relatively poor functionalization rate. The average concentration of the carboxylic group was 38.4 nmol m3, which is equivalent to 8.4 mmol mol1 of OC (w1 carbon out of 118 is linked to a carboxylic group). In comparison, carboxylic average contents of 20.8 mmol mol1 of OC (w1 carbon out of 48 is linked to a carboxylic group) were measured for ambient aerosol collected at an urban background site, in winter (Dron, 2008). The MS fragments corresponding to the R-COOH functional group in PM2.5 are mainly observed in the m/z range 200–300 amu (Fig. 3). However, peaks over 300 amu were also observed, which suggest the presence of high molecular weight acids in the aerosol. HULIS are

6194

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

outside the tunnel through radical-initiated reactions of OC compounds (Reisen and Arey, 2005). Finally, the PM2.5 carbonyl group concentration was between 10.7 and 12.7 nmol m3 and accounts for 2.4 mmol mol1 of OC.

3.3. Organic trace compounds

Fig. 3. Mass spectrum obtained in NL32 mode for derivatized carboxylic acids in the mass range of m/z 120–600 uma, from a PM2.5 sample.

a fraction of these higher mass acids, the concentration measured represent 18–25% of WSOC, and 2–5% of OC measured inside the tunnel. Considering an OM/OC ratio of 1.81 (Salma et al., 2007), average HULIS fraction is roughly estimated to account for 5% of the OM (Fig. 2). This is a relatively small fraction compared to the fractions measured in other environments, as HULIS accounted for 11–13% of OC in European continental sites (Baduel et al., 2009). High loadings of nitro functional group were measured in the vehicles-emitted organic aerosol. The average concentration of the PM2.5 nitro group was 37.1 nmol m3, equivalent to 8.3 mmol mol1 of OC (w1 carbon out of 127 is linked to a nitro group). These contributions were clearly higher than those previously measured in ambient atmospheric aerosol at urban background sites (<0.4 mmol mol1 of OC) (Dron et al., 2008a). The nitro group found in aerosol from the tunnel can be partially attributed to nitroPAH compounds, previously identified in the diesel exhaust PM (Heeb et al., 2008), but a large fraction can also be formed inside or

The chromatographically elutable organic compounds can be subdivided into resolved and unresolved complex mixture (UCM) that appears, in the GC/MS chromatograms, as a wide hump underlying the resolved peaks of individual compounds (Fig. S1, Supplementary material). The unresolved organic mass (UCM) was nearly 8 times greater than the resolved mass (R) (UCM:Rw8). The predominance of the UCM fraction in the vehicular emissions comparing to the resolved fraction is consistent with Schauer et al. (1999) findings. The identified and quantified individual organic compound accounts for about 4% of the OM (Fig. 2). The relative contributions of the different organic compounds to OC give comparable results for PM2.5 and PM10, with only little differences for some compound classes. The following discussion focuses mainly on the PM2.5 fraction, allowing direct comparison with published data. N-alkanes (C18–C34) are the most abundant fraction of the total quantified organic compound mass representing 2.7% of OM and 0.7% of PM2.5 (Table 2). Similar contributions to OM were reported by Rogge et al. (1993a) for diesel-powered engines (2.3% of the OM). In contrast, catalyst gasoline powered engines yield lower levels of n-alkanes (1.0% of the OM) (Rogge et al., 1993a). Fig. 4a shows the distributions of the n-alkanes normalized to OC (PM2.5 and PM10) from this study together with those for gasoline and diesel emissions (Rogge et al., 1993a). These profiles are dominated by C20-C21 alkanes. The lower molecular weight (MW) n-alkanes (C27) contributions measured here (w1400 mg g1 of OC) cannot be entirely attributed to the tailpipe sources that yield relatively low amounts of these components (w210 mg g1 of OC) (Rogge et al., 1993a). Potential non-tailpipe high emitters of these alkanes are tire wear dust (w2500 mg g1 of OC) and the re-suspension of road materials (w810 mg g1 of OC)

Table 2 n-Alkane, branched alkane and polycyclic aromatic hydrocarbon concentrations in the traffic related primary PM2.5 and PM10 (average (min–max)).

n-Alkanes [ng m3] n-Octadecane (A18) n-Nonadecane (A19) n-Eicosane (A20) n-Heneicosane (A21) n-Docosane (A22) n-Tricosane (A23) n-Tetracosane (A24) n-Pentacosane (A25) n-Hexacosane (A26)

PM2.5

PM10

Note

63.3 73.8 167 230 184 143 130 78.0 83.4

158 179 422 224 256 195 197 150 117

b, a, b, b, a, a, > b, > a, > b, >

(47.7–78.1) (61.3–86.3) (138–195) (218–241) (162–205) (120–166) (105–156) (77.2–78.8) (82.2–84.7)

(141–170) (171–187) (353–491) (205–242) (230–283) (172–219) (168–226) (120–179) (97.3–138)

PM2.5

PM10

n-Heptacosane (A27) n-Octacosane (A28) n-Nonacosane (A29) n-Triacontane (A30) n-Hentriacontane (A31) n-Dotriacontane (A32) n-Tritriacontane (A33) n-Tetratriacontane (A34)

88.2 83.6 121 94.6 103 96.1 80.9 79.4

96.5 40.2 106 nd 57.7 nd 22.3 nd

34.6 (31.5–37.7)

Branched alkanes [ng m3] Pristane 38.5 (32.4–44.6)

78.6 (75.8–81.4)

a, -

Phytane

Polycyclic aromatic hydrocarbons [ng m3] Fluoranthene 45.0 (40.6–49.5) Pyrene 23.3 (21.7–25.0) Triphenylene 21.7 (21.0–22.6) Benzo[a]anthracene 1.84 (1.75–1.92) Chrysene 1.96 (1.02–2.93) Benzo[e]pyrene (BeP) nd

64 31.4 26.4 2.45 2.83 nd

a, a, a, a, a, a,

Benzo[b]fluoranthene Benzo[k]fluoranthene Benzo[a]pyrene Benzo[ghi]perylene Indeno(1,2,3-cd)pyrene (IP) Coronene

(58.5–69.6) (29.3–33.6) (25.1–27.7) (2.15–2.75) (2.56–3.11)

< > > < -

(72.0–104) (64.7–102) (106–135) (80.6–109) (83.3–122) (73.6–118) (78.7–83.1) (74.1–84.8)

nd 0.019 (0.009–0.031) 2.42 (2.35–2.48) 0.130 (0.130–0.140) nd nd

Note (71.8–121) (19.3–61.1) (101–112) (26.1–89.3) (9.63–34.6)

80.7 (66.6–94.9) 0.36 0.192 6.73 0.432 nd 0.161

(0.230–0.490) (0.130–0.250) (5.08–8.38) (0.290–0.570) (0.060–0.260)

b, < a, < b, < a, < b, < b, b, a, -

a, a, a, a, a, a, a,

> > > > < -

(a and b) Identification notes: the quantification of the organic compounds is based on the response factors of a – authentic standards, b – average of alkanes with the closer carbon number. (-, < and >) Sampling artefacts notes: organic compounds for which sampling artefacts induce errors higher (>), lower (<) than 30% according to Goriaux et al. (2006) and Sihabut et al. (2005) or not measured before (-).

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

Fig. 4. Comparison of the PM2.5 and PM10 individual organic compounds distributions ((a) n-alkanes and (b) hopanes) within the Tunnel (grey bars) with the emission rates of each compound from gasoline vehicles (dotes) and diesel vehicles (triangles) determined by Rogge et al. (1993a,b) for n-alkanes and Zielinska et al. (2004) for hopanes.

(Rogge et al., 1993b). It is also interesting to note that for PM10 samples, high MW n-alkane concentrations are much lower than those for PM2.5 samples. This difference highlights the influence of the washing procedure of the tunnel (achieved between the PM10 and PM2.5 sampling) on re-suspension processes conditions and in particular on the amount of materials available for re-suspension.

6195

These results also show that a significant fraction of these n-alkanes is related to the non-tailpipe sources even for particle less than 2.5 mm in diameter. Two branched alkanes, proposed as organic markers for fuel utilisation, were also quantified: pristane (C19) and phytane (C20). Average concentrations of pristane and phytane in PM2.5 were 38.5 and 34.6 ng m3, respectively. The total PAH concentration measured in PM2.5 fraction was 95.4 ng m3, accounting for 0.13% of the OM mass (Table 2). Since these compounds are considered as tracers of combustion sources, many studies have been carried out to characterize their emission profiles and to point out diagnostic ratios specific to vehicular emissions (Miguel et al., 1998; Schauer et al., 1999, 2002; Zielinska et al., 2004). They suggested that diesel-vehicular emissions are enriched in low MW PAH (4 cycles), whereas, high MW PAH are more abundant in gasoline engine emissions. In our study, the most abundant PAH was fluoranthene (45.0 ng m3) followed by pyrene (23.3 ng m3). In contrast, indeno(1,2,3-cd)pyrene, often used in source apportionment studies as organic marker for emissions from gasoline exhausts, was not detected. Among the high MW PAHs, benzo(a)pyrene was the most abundant (2.42 ng m3). The PAH distribution found here points towards a much higher contribution from diesel emissions compared to gasoline emissions. Petroleum biomarkers including diasteranes, steranes and hopanes classes are fossil compounds present in crude oils, used to trace the petrogenic inputs in the environment, particularly in the airborne PM (Rogge et al., 1993a; Schauer et al., 1999, 2002). Because these markers belong to the higher boiling point fraction, they are not found in gasoline or diesel fuel but instead originate from the unburned lubricating oils and are found in both types of engines emissions. In this study, sterane total concentration was 47.5 ng m3, contributing to 0.067% of the OM mass (Table 3). The sum of the average concentration of hopanes was 59.0 ng m3, accounting for 0.08% of the OM mass (Table 3). hopane series maximizes at C29 and presents the typical 22SjR pairs of extended hopane homologues (C31), with S epimers concentrations always higher than those of the R epimers (S/S þ R ¼ 0.65), which is characteristic of vehicle exhaust particles (Simoneit, 1984). The hopane concentrations measured in this study were compared with several published profiles (Rogge et al., 1993a; Schauer et al., 1999, 2002; Zielinska et al., 2004) in term of distribution patterns

Table 3 Petroleum biomarkers concentrations in the traffic related primary PM2.5 and PM10 (average (min–max)). PM2.5 Diasteranes [ng m3] 20S,13b(H), 17a(H)-diacholestane Steranes [ng m3] 20R>S,5a(H),14b(H), 17b(H)-cholestane 20R,5a(H),14a(H), 17a(H)-cholestane 20R&S,5a(H),14b(H), 17b(H)-ergostane Hopanes [ng m3] Trisnorneohopane (TS) 17a(H)-trisnorhopane (Tm) 17a(H),21b(H)-norhopane (H29) 17a(H),21b(H)-hopane (H30) 22S,17a(H),21b(H)homohopane (S-H31)

PM10

Note

PM2.5

PM10

Note

20R,13b(H),17a(H)-diacholestane

3.21 (3.15–3.28) 3.73 (3.52–3.96) b, -

8.64 (7.65–9.62) 10.8 (10.1–11.4) b, -

20S,5a(H),14a(H), 17a(H)-stigmastane

4.67 (4.31–5.04) 5.04 (4.95–5.13) b, -

9.36 (8.47–10.3) 12.4 (11.7–13.1) a, -

20R&S,5a(H),14b(H), 17b(H)-stigmastane

12.5 (11.2–13.9) 14.3 (14.2–14.4) b, -

7.31 (6.44–8.19) 11.1 (10.8–11.3) b, -

20S,5a(H),14a(H), 17a(H)-stigmastane

5.02 (4.47–5.57) 5.79 (4.94–6.64) b, -

5.39 4.68 16.3 12.2 6.84

5.65 (5.15–6.15) 7.08 (6.77–7.39) c, < 22R,17a(H),21b(H)-homohopane (R–H31) 22S,17a(H),21b(H)-bishomohopane (S-H32) 3.59 (2.86–4.31) 5.04 (4.52–5.57) c, < 22R,17a(H)-21b(H)-bishomohopane (R–H32) 0.865 (0.86–0.87) nd c, < 17a(H)-21b(H)-22S-trishomohopane (S-H33) 2.34 (1.96–2.72) 4.43 (4.27–4.59) c, < 17a(H)-21b(H)-22R-trishomohopane (R–H33) 1.05 (0.95–1.16) nd c, <

4.9 (4.50–5.30) 5.69 (5.40–5.98) b, -

(4.89–5.89) (4.48–4.88) (14.9–17.8) (10.7–13.7) (6.16–7.53)

7.84 6.57 21.2 16.9 8.62

(7.37–8.32) (6.18–6.95) (20.5–21.9) (15.3–18.7) (7.23–10.0)

c, < c, < c, < a, < c, <

(a, b and c) Identification notes: the quantification of the organic compounds is based on the response factors of a – authentic standards, b – 20R,5a(H),14a(H),17a(H)cholestane and c – 17a(H),21b(H)-hopane. (-, < and >) Sampling artefacts notes: organic compounds for which sampling artefacts induce errors higher (>), lower (<) than 30% according to Goriaux et al. (2006) and Sihabut et al. (2005) or not measured before (-).

6196

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

Table 4 Phthalate ester and carboxylic acid concentrations in the traffic related primary PM2.5 and PM10 (average (min–max)). PM2.5 Phthalates esters [ng m3] Di-ethyl-phthalate Di-isobutyl-phthalate (DiBP) Di-nbutyl-phthalate (DBP)

5.38 (4.79–5.97) 31.5 (31.5–31.6) 16.5 (14.7–18.2)

n-Alkanoic acids [ng m3] Decanoic acid Undecanoic acid Dodecanoic acid Tridecanoic acid Tetradecanoic acid Pentadecanoic acid Hexadecanoic acid Heptadecanoic acid

6.46 2.93 9.67 4.08 13.3 6.62 80.1 5.38

PM10

Note

10.9 (10.5–11.3) 46.3 (45.9–46.7) 27.7 (27.5–27.8)

PM2.5

PM10

Note

a, b, a, -

Nbutyl-benzyl-phtalate Bis(2ethylhexyl)phthalate (DEHP)

11.4 (11.3–11.5) 37.9 (35.5–40.5)

23.3 (22.6–24.1) 48.6 (48.3–48.8)

a, a, -

120 7.55 13.1 nd nd nd 13.9

a, < c, a, < c, a, < c, a, -

(6.23–6.68) (2.69–3.18) (8.6–10.7) (3.18–4.98) (12.6–14.0) (6.00–7.25) (66.1–87.9) (3.97–6.8)

8.99 4.49 11.3 6.76 13.2 12.1 127.5 7.61

(8.38–9.61) (3.83–5.14) (9.81–12.8) (6.15–7.36) (13.1–13.2) (8.2–16.0) (93.0–162) (6.46–8.76)

a, > c, > a, > c, > c, > a, > a, < a, <

Octadecanoic acid Nonadecanoic acid Eicosanoic acid Heneicosanoic acid Docosanoic acid Tricosanoic acid Tetracosanoic acid

63.1 4.03 5.87 nd nd nd 14.1

Aromatic and oxalic acid [ng m3] 2-Methylbenzoic acid 0.35 (0.29–0.41) 2-Formylbenzoic acid 2.51 (1.96–3.03) 1,2-Benzendicarboxylic acid nd 1,3-Benzendicarboxylic acid 1.01 (0.94–1.08) 1,4-Benzendicarboxylic acid 3.89 (3.75–4.04)

1.51 3.63 nd 1.05 4.54

(1.36–1.66) (3.25–4.02)

a, a, a, a, a,

4-Methyl-1,2-benzendicarboxylic acid 1,2,3-Benzentricarboxylic acid 1,2,4-Benzentricarboxylic acid 1,3,5-Benzentricarboxylic acid Oxalic acid

nd nd nd nd 234 (194–274)

(0.87–1.24) (4.16–4.92)

-

(46.6–79.5) (3.60–4.46) (5.47–6.26)

(5.0–22.8)

(104–138) (7.52–7.58) (12.6–13.7)

(11.6–16.3)

nd nd nd nd 249 (203–364)

a, a, a, a, a,

-

(a, b and c) Identification notes: the quantification of the organic compounds is based on the response factors of a – authentic standards, b – di-nbutyl-phthalate and c – the nalkanoic acid with the closest carbon number. (-, < and >) Sampling artefacts notes: organic compounds for which sampling artefacts induce errors higher (>), lower (<) than 30% according to Goriaux et al. (2006) and Sihabut et al. (2005) or not measured before (-).

and relative contributions to OC (Fig. 4b). These characteristics are mainly inherited from the crude oils from which the lubricating oil was manufactured and can also depend of the engine type, since the catalyst systems induce the destruction of hopanes before they reach the tailpipes yielding lower contributions to OC (Schauer et al., 2002). The hopanes distribution pattern found in our study confirms the profiles recently established by Zielinska et al. from US vehicular emissions (Zielinska et al., 2004). Hopanes to OC contributions comparable to our findings (0.049% and 0.124% for gasoline and diesel emissions, respectively) are also reported in this last study (Zielinska et al., 2004). Five Phthalate esters, dominated by DEHP, DiBP and DBP, were also detected in the aerosol from the tunnel at a total concentration of 165 ng m3, accounting for approximately 0.23% of the OM mass (Table 4). To the best of our knowledge, phthalates have never been reported in direct emissions from vehicle exhausts and their presence in the PM is most probably associated with non-exhaust sources. Phthalate esters are widely used as plasticizers in several polymeric materials, including construction materials, paint pigments, caulk, adhesives, and lubricants (Staples et al., 1997). Their evaporation from the polymeric matrixes, including coating materials and polymers present in the tunnel construction materials and in the vehicle fleet, can be a potential source of these additives. Several carboxylic acids classes were also determined in PM2.5 from the tunnel, including linear alkanoic acids (212 ng m3), aromatic acids (7.75 ng m3) and oxalic acid (234 ng m3) (Table 4). These carboxylic acids can result from the combustion processes of the fuel. Oxalic acid is usually associated with the secondary fraction of the organic aerosol (Wang et al., 2006), but several studies provide evidence of the oxalic acid emissions from primary processes including vehicular emissions (Wang et al., 2006). However, we cannot exclude an influence of Marseille background particles to the concentration of oxalic acid observed inside the tunnel. The sum of the identified carboxylic acids accounts for 6.1 nmol of R-COOH m3, which is 15.1% of the total RCOOH (38.4 nmol m3; Section 3.2). The unidentified fraction of R-COOH can be associated with other dicarboxylic acids and poly-carboxylic acids that contribute to the HULIS fraction.

3.4. Implication for CMB modeling applications Fig. 5 compares the organic profile obtained in this study to those reported by Schauer et al. (1999) for a medium duty diesel truck emissions and Schauer et al. (2002) for catalyst equipped gasoline vehicles. The organic species and the vehicular emission profiles chosen are considered in the most recent source apportionment studies using CMB modeling (Ke et al., 2007; Sheesley et al., 2007). This comparison clearly shows that the profile determined in this study is very similar to the diesel profile (Schauer et al., 1999). This highlights two major points: (i) it suggests that vehicular emissions profiles commonly used in the literature for CMB modeling are rather homogeneous over large geographic areas and could be used when specific vehicular emission profiles are not available. (ii) It indicates the high predominance of the diesel emissions compared to those of catalyst equipped gasoline vehicles, although the latter accounts for the half of the French vehicle fleet. The predominance of the diesel emissions found here can be explained by the much higher emission rates of PM from the

Fig. 5. Comparison, in terms of organic tracers to OC contributions, between the PM2.5 organic profile reported in this study (open circles) and those reported by Schauer et al., 1999 for a medium duty diesel truck emissions (thin crosses) and Schauer et al. (2002) for catalyst equipped gasoline vehicles (open triangles). For the species that are not found in the emissions their detection limits are considered.

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

diesel vehicles (w190 mg km1) compared to those from catalyst equipped gasoline vehicles (w7 mg km1) (Schauer et al., 1999, 2002). This highlights the difference between the overall vehicular emissions in France and in the US, where the diesel emissions account roughly for the half of the total vehicular emissions (Gertler, 2005). The source profiles established here are believed to reflect, for the compounds taken into consideration, the PM emissions from an overall vehicular activity in France (dominated by diesel emissions), and thus may be well appropriate for source apportionment modeling in this region. 4. Conclusion In this study, the traffic related primary aerosol characteristic of French light-duty vehicular fleet was characterized. The organic source profiles established here are in good agreement with the profiles previously established in the U.S. using dynamometer facilities or tunnel environments. This suggests that these source profiles can be considered homogeneous over large geographic areas and thus could be used for source apportionment purposes when specific vehicular emission profiles are not available. The aerosol chemical fingerprint is characterized by high concentrations of EC relative to OC (EC/OC ¼ 1.8) and very low concentrations of the higher molecular weight PAH, which points towards a much higher contribution from diesel exhausts comparing to gasoline exhausts. When applying such source profile in CMB modeling studies, it is necessary to keep in mind the potential modification of the organic profile from the source to the receptor site mainly related to gas/particle partitioning of the organic materiel occurring during the dilution process as well as chemical aging of the organic markers which is not well understood yet (Robinson et al., 2007). Finally, high concentrations of highly toxic organic compounds, such as benzo(a)pyrene (2–9 ng m3) and phthalates esters (165 ng m3) associated with high concentrations of submicrometer particles (ranging from 6.105 to 106 cm3) have been observed. This can acute health outcomes to those that have to commute regularly through this tunnel (average of 40 000 vehicles per day). Acknowledgments This work has been supported by MEDAD (Ministe`re de l’Ecologie, du De´veloppement et de l’Ame´nagement Durables), ADEME (Agence de l’Environnement et de la Maıˆtrise de l’Energie) through the research program PRIMEQUAL (project FORMES) and INSU (Institut national des sciences de l’Univers). The authors also acknowledge the personal staff of the society in charge of the tunnel for their help. Appendix. Supplementary material Supplementary data associated with this article can be found in the online version at doi:10.1016/j.atmosenv.2009.09.001. References 2008/50/EC, 2008. d.: directive 2008/50/EC of the European parliament and of the council of 21 May 2008 on ambient air quality and cleaner air for Europe. Official Journal of the European Union 1–44. Arthaut, R., 2005. Le budget transports des me´nages depuis 40 ans. Institut national de la statistique et de l’e´tude e´conomique (INSEE). Aymoz, G., Jaffrezo, J.L., Chapuis, D., Cozic, J., Maenhaut, W., 2007. Seasonal variation of PM10 main constituents in two valleys of the French Alps. I: EC/OC fractions. Atmospheric Chemistry and Physics 7, 661–675.

6197

Baduel, C., Voisin, D., Jaffrezo, J.L., 2009. Comparison of analytical methods for HULIS measurements in atmospheric particles. Atmospheric Chemistry and Physics 9, 5949–5962. Besombes, J.L., Maitre, A., Patissier, O., Marchand, N., Chevron, N., Stoklov, M., Masclet, P., 2001. Particulate PAHs observed in the surrounding of a municipal incinerator. Atmospheric Environment 35, 6093–6104. Cavalli, F., Viana, M., Yttri, K.E., Genberg, J., Putaud, J.-P. Toward a standardised thermal-optical protocol for measuring atmospheric organic and elemental carbon: the EUSAAR protocol. Atmospheric Measurement Techniques Discussion, in press. Chow, J.C., Watson, J.G., Crow, D., Lowenthal, D.H., Merrifield, T., 2001. Comparison of IMPROVE and NIOSH carbon measurements. Aerosol Science and Technology 1, 23–34. DRI, 2003. Analysis of Semi-volatile Organic Compound by GC/MS. Desert Research Institute, DRI Standard Operating Procedure., 1–25. Dron, J., 2008. Analyse fonctionnelle par spectrome´trie de masse tandem: application aux ae´rosols organiques atmosphe´riques, PhD thesis (universite´ de provence). Dron, J., Eyglunent, G., Temime-Roussel, B., Marchand, N., Wortham, H., 2007. Carboxylic acid functional group analysis using constant neutral loss scanningmass spectrometry. Analytica Chimica Acta 1, 61–69. Dron, J., Abidi, E., El Haddad, I., Marchand, N., Wortham, H., 2008a. Precursor ion scanning-mass spectrometry for the determination of nitro functional groups in atmospheric particulate organic matter. Analytica Chimica Acta 2, 184–195. Dron, J., Zheng, W., Marchand, N., Wortham, H., 2008b. New method to determine the total carbonyl functional group content in extractable particulate organic matter by tandem mass spectrometry. Journal of Mass Spectrometry 8, 1089–1098. Gertler, A.W., 2005. Diesel vs. gasoline emissions: does PM from diesel or gasoline vehicles dominate in the US? Atmospheric Environment 13, 2349–2355. Goriaux, M., Jourdain, B., Temime, B., Besombes, J.L., Marchand, N., Albinet, A., LeozGarziandia, E., Wortham, H., 2006. Field comparison of particulate PAH measurements using a low-flow denuder device and conventional sampling systems. Environmental Science & Technology 20, 6398–6404. Grahame, T.J., Schlesinger, R.B., 2007. Health effects of airborne particulate matter: do we know enough to consider regulating specific particle types or sources? Inhalation Toxicology 6–7, 457–481. Heeb, N.V., Schmid, P., Kohler, M., Gujer, E., Zennegg, M., Wenger, D., Wichser, A., Ulrich, A., Gfeller, U., Honegger, P., Zeyer, K., Emmenegger, L., Petermann, J.L., Czerwinski, J., Mosimann, T., Kasper, M., Mayer, A., 2008. Secondary effects of catalytic diesel particulate filters: conversion of PAHs versus formation of NitroPAHs. Environmental Science & Technology 10, 3773–3779. Jaffrezo, J.-L., Aymoz, G., Delaval, C., Cozic, J., 2005. Seasonal variations of the water soluble organic carbon mass fraction of aerosol in two valleys of the French Alps. Atmospheric Chemistry and Physics 10, 2809–2821. Jamriska, M., Morawska, L., Thomas, S., He, C., 2004. Diesel bus emissions measured in a tunnel study. Environmental Science & Technology 24, 6701–6709. Jaffrezo, J.L., Calas, T., Bouchet, M., 1998. Carboxylic acids measurements with ionic chromatography. Atmospheric Environment 14–15, 2705–2708. Ke, L., Ding, X., Tanner, R.L., Schauer, J.J., Zheng, M., 2007. Source contributions to carbonaceous aerosols in the Tennessee Valley Region. Atmospheric Environment 39, 8898–8923. Mader, B.T., Pankow, J.F., 2001. Gas/solid partitioning of semivolatile organic compounds (SOCs) to air filters. 3. An analysis of gas adsorption artifacts in measurements of atmospheric SOCs and organic carbon (OC) when using Teflon membrane filters and quartz fiber filters. Environmental Science & Technology 17, 3422–3432. Miguel, A.H., Kirchstetter, T.W., Harley, R.A., Hering, S.V., 1998. On-road emissions of particulate polycyclic aromatic hydrocarbons and black carbon from gasoline and diesel vehicles. Environmental Science & Technology 4, 450–455. Ntziachristos, L., Ning, Z., Geller, M.D., Sheesley, R.J., Schauer, J.J., Sioutas, C., 2007. Fine, ultrafine and nanoparticle trace element compositions near a major freeway with a high heavy-duty diesel fraction. Atmospheric Environment 27, 5684–5696. Querol, X., Alastuey, A., Ruiz, C.R., Artinano, B., Hansson, H.C., Harrison, R.M., Buringh, E., ten Brink, H.M., Lutz, M., Bruckmann, P., Straehl, P., Schneider, J., 2004. Speciation and origin of PM10 and PM2.5 in selected European cities. Atmospheric Environment 38, 6547–6555. Reisen, F., Arey, J., 2005. Atmospheric reactions influence seasonal PAH and nitroPAH concentrations in the Los Angeles basin. Environmental Science & Technology 1, 64–73. Robinson, A.L., Donahue, N.M., Shrivastava, M.K., Weitkamp, E.A., Sage, A.M., Grieshop, A.P., Lane, T.E., Pierce, J.R., Pandis, S.N., 2007. Rethinking organic aerosols: semivolatile emissions and photochemical aging. Science 5816, 1259–1262. Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993a. Sources of fine organic aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environmental Science & Technology 4, 636–651. Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993b. Sources of fine organic aerosol. 3. Road dust, tire debris, and organometallic brake lining dust – roads as sources and Sinks. Environmental Science & Technology 9, 1892–1904.

6198

I. El Haddad et al. / Atmospheric Environment 43 (2009) 6190–6198

Salma, I., Ocskay, R., Chi, X.G., Maenhaut, W., 2007. Sampling artefacts, concentration and chemical composition of fine water-soluble organic carbon and humiclike substances in a continental urban atmospheric environment. Atmospheric Environment 19, 4106–4118. Sannigrahi, P., Sullivan, A.P., Weber, R.J., Ingall, E.D., 2006. Characterization of watersoluble organic carbon in urban atmospheric aerosols using solid-state C-13 NMR spectroscopy. Environmental Science & Technology 3, 666–672. Schauer, J.J., Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1996. Source apportionment of airborne particulate matter using organic compounds as tracers. Atmospheric Environment 22, 3837–3855. Schauer, J.J., Kleeman, M., Cass, G., Simoneit, B.R.T., 1999. Measurement of emissions from air pollution sources. 2. C1 through C30 organic compounds from medium duty diesel trucks. Environmental Science & Technology 10, 1578–1587. Schauer, J.J., Kleeman, M.J., Cass, G.R., Simoneit, B.R.T., 2002. Measurement of emissions from air pollution sources. 5. C1-C32 organic compounds from gasolinepowered motor vehicles. Environmental Science & Technology 6, 1169–1180. Sheesley, R.J., Schauer, J.J., Zheng, M., Wang, B., 2007. Sensitivity of molecular marker-based CMB models to biomass burning source profiles. Atmospheric Environment 39, 9050–9063.

Sihabut, T., Ray, J., Northcross, A., McDow, S.R., 2005. Sampling artifact estimates for alkanes, hopanes, and aliphatic carboxylic acids. Atmospheric Environment 37, 6945–6956. Simoneit, B.R.T., 1984. Organic matter of the troposphere. III. Characterization and sources of petroleum and pyrogenic residues in aerosols over the western United States. Atmospheric Environment 1, 51–67. Staples, C.A., Peterson, D.R., Parkerton, T.F., Adams, W.J., 1997. The environmental fate of phthalate esters: a literature review. Chemosphere 4, 667–749. Turpin, B.J., Lim, H.-J., 2001. Species contributions to PM2.5 mass concentrations: revisiting common assumptions for estimating organic mass. Aerosol Science and Technology 1, 602–610. Wang, H.B., Kawamura, K., Ho, K.F., Lee, S.C., 2006. Low molecular weight dicarboxylic acids, ketoacids, and dicarbonyls in the fine particles from a roadway tunnel: possible secondary production from the precursors. Environmental Science & Technology 20, 6255–6260. Zielinska, B., Sagebiel, J., McDonald, J.D., Whitney, K., Lawson, D.R., 2004. Emission rates and comparative chemical composition from selected in-use diesel and gasoline-fueled vehicles. Journal of the Air & Waste Management Association 9, 1138–1150.